Daily River Flow Forecasting Using Artificial Neural Networks and Auto-Regressive Models


Abstract: Estimating the flows of rivers can have a significant economic impact, as this can help in agricultural water management and in providing protection from water shortages and possible flood damage. This paper provides forecasting benchmarks for river flow prediction in the form of a numerical and graphical comparison between neural networks and auto-regressive (AR) models. Benchmarking was based on 7 and 4-year periods of continuous river flow data for 2 rivers in the USA, the Blackwater River and the Gila River, and a 2-year period of streamflow data for the Filyos Stream in Turkey. The choice of appropriate artificial neural network (ANN) architectures for hydrological forecasting, in terms of hidden layers and nodes, was investigated. Three simple neural network (NN) architectures were then selected for comparison with the AR model forecasts. Sum of square errors (SSEs) and correlation statistic measures were used to evaluate the models' performances. The benchmark results showed that NNs were able to produce better results than AR models when given the same data inputs.

Keywords: Streamflow forecasting, Neural networks, Auto-regressive models

Full Text: PDF